Gaining meaningful insights into bacterial communities associated with animal hosts requires the provision of high-quality nucleic acids. Although many studies have compared DNA extraction methods for samples with low bacterial biomass (e.g. water) or specific PCR inhibitors (e.g. plants), DNA extraction bias in samples without inherent technical constraint (e.g. animal samples) has received little attention. Furthermore, there is an urgent need to identify a DNA extraction methods in a high-throughput format that decreases the cost and time for processing large numbers of samples. We here evaluated five DNA extraction protocols, using silica membrane-based spin columns and a 96-well microplate format and based on either mechanical or enzymatic lysis or a combination of both, using three bacterial mock communities and Illumina sequencing of the V4 region of the 16SrRNA gene. Our results showed that none of the DNA extraction methods fully eliminated bias associated with unequal lysis efficiencies. However, we identified a DNA extraction method with a lower bias for each mock community standard. Of these methods, those including an enzymatic lysis showed biases specific to some bacteria. Altogether, these results again demonstrate the importance of DNA extraction standardization to be able to compare the microbiome results of different samples. In this attempt, we advise for the use of the 96-well DNeasy Blood and Tissue kit (Qiagen) with a zirconia bead-beating procedure, which optimizes altogether the cost, handling time and bacteria-specific effects associated with enzymatic lysis.